Speaker Recognition and Diarization

This chapter presents a continuously growing field that promises a wealth of applications far beyond the field of speech processing: the automatic identification of persons from their uttered speech. Research is currently focusing mainly on two tasks: The task of speaker detection is to verify the identity of a new speaker against a set of pretrained speaker models. The task of speaker diarization is to find speech segments of the same speaker without any a priori knowledge. The chapter introduces the general ideas in the two fields then it continues to explain the task of speaker diarization by providing an overview of current work before providing a more detailed description of a concrete example of a diarization system. Then, variants and current research topics are discussed. It presents speaker recognition in a similar way. Finally it concludes the chapter pointing to open problems. Controlled Vocabulary Terms speaker recognition